Sorry, you need to enable JavaScript to visit this website.

facebooktwittermailshare

Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication

Abstract: 

For lossy image compression, we develop a neural-based system which learns a nonlinear estimator for decoding from quantized representations. The system links two recurrent networks that \help" each other reconstruct same target image patches using complementary portions of spatial context that communicate via gradient signals. This dual agent system builds upon prior work that proposed the iterative refinement algorithm for recurrent neural network (RNN)based decoding which improved image reconstruction compared to standard decoding techniques. Our approach, which works with any encoder, neural or non-neural, This system progressively reduces image patch reconstruction error over a fixed number of steps. Experiment with variants of RNN memory cells, with and without future information, find that our model consistently creates lower distortion images of higher perceptual quality compared to other approaches. Specifically, on the Kodak Lossless True Color Image Suite, we observe as much as a 1:64 decibel (dB) gain over JPEG, a 1:46 dB gain over JPEG 2000, a 1:34 dB gain over the GOOG neural baseline, 0:36 over E2E (a modern competitive neural compression model), and 0:37 over a single iterative neural decoder.

up
0 users have voted:

Paper Details

Authors:
Ankur Mali, Alexander G. Ororbia, C. Lee Giles
Submitted On:
31 March 2020 - 4:39pm
Short Link:
Type:
Presentation Slides
Event:
Presenter's Name:
Ankur Mali
Paper Code:
173
Session:
Session 1
Document Year:
2020
Cite

Document Files

DCC_2020_Ankur Mali.pptx

(37)

Subscribe

[1] Ankur Mali, Alexander G. Ororbia, C. Lee Giles , "Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication", IEEE SigPort, 2020. [Online]. Available: http://sigport.org/5086. Accessed: Jul. 10, 2020.
@article{5086-20,
url = {http://sigport.org/5086},
author = { Ankur Mali; Alexander G. Ororbia; C. Lee Giles },
publisher = {IEEE SigPort},
title = {Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication},
year = {2020} }
TY - EJOUR
T1 - Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication
AU - Ankur Mali; Alexander G. Ororbia; C. Lee Giles
PY - 2020
PB - IEEE SigPort
UR - http://sigport.org/5086
ER -
Ankur Mali, Alexander G. Ororbia, C. Lee Giles . (2020). Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication. IEEE SigPort. http://sigport.org/5086
Ankur Mali, Alexander G. Ororbia, C. Lee Giles , 2020. Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication. Available at: http://sigport.org/5086.
Ankur Mali, Alexander G. Ororbia, C. Lee Giles . (2020). "Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication." Web.
1. Ankur Mali, Alexander G. Ororbia, C. Lee Giles . Sibling Neural Estimators: Improving Iterative Image Decoding with Gradient Communication [Internet]. IEEE SigPort; 2020. Available from : http://sigport.org/5086